Z. Franco, Q.D. Nguyen: Sciencedirect
Z. Franco, Q.D. Nguyen: Sciencedirect
Z. Franco, Q.D. Nguyen: Sciencedirect
Fuel
journal homepage: www.elsevier.com/locate/fuel
a r t i c l e
i n f o
Article history:
Received 18 June 2009
Received in revised form 16 September
2010
Accepted 22 September 2010
Available online 16 October 2010
Keywords:
Vegetable oil
Diesel fuel
Diesel blend
Viscosity
Blending rule
a b s t r a c t
Straight vegetable oils provide cleaner burning and renewable alternatives to diesel fuel, but their inherently high viscosity compared to petroleum based diesel is undesirable for diesel engines. Lowering the
viscosity can be simply achieved by either increasing the temperature of the oil or by blending it with
diesel fuel, or both. In this work the rheological properties of diesel fuel and vegetable oil mixtures at different compositions were studied as a function of temperature to determine a viscositytemperature
composition relationship for use in design and optimization of heating and fuel injection systems used
in diesel engines. The vegetable oils used were corn, canola, olive, peanut, soybean and sunower oils
which are of commercial food grade. All the vegetable oils and their blends with No. 2 diesel fuel showed
time-independent Newtonian behaviour within the test temperatures between 20 C and 80 C. Viscosities of the pure oils and diesel were satisfactorily correlated with temperature by means of the Arrhenius
typed relationship. The Arrhenius blending rule was found applicable to describing the composition
dependence of viscosity all vegetable oilsdiesel blends at a xed temperature. These relations were combined to develop a simple mixture viscosity model to predict the viscosity of the vegetable oildiesel
blends as functions of temperature and composition based on properties of the pure components.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
With the instability in supply and price of crude oil in the world
market and the need to reduce exhaust emission, there is a renewed interest in the search for alternative sources of transport
fuel. Vegetable oils represent a ready, renewable and clean energy
source that has shown promise as a substitute to petroleum diesel
fuel for diesel engines [13]. There are different ways of utilising
vegetable oils as a substitute for petroleum diesel: (i) crude or rened neat oil; (ii) mixture of oil with diesel fuel with appropriate
dilutions; and (iii) alkyl ester products from the transesterication
process [35]. The advantages of vegetable oils, as compared to
other renewable energy sources, are that they can be produced
from numerous oil seed crops that can be cultivated anywhere,
have high energy contents, and exhibit clean combustion behaviour with zero lifecycle CO2 emission and negligible SO2 generation. Due to their low volatility and high ash points, vegetable
oils can be handled and stored with minimal safety precaution.
The greatest and immediate potential of vegetable oils is in agriculture, where they can be used as a sustainable source of fuel to supply some or all the energy needed to operate farm machinery and
equipment.
A major problem concerning the use of straight vegetable oils in
diesel engines is their inherently high viscosity, which causes poor
Corresponding author. Tel.: +61 8 8303 5456; fax: +61 8 8303 4373.
E-mail address: dzuy.nguyen@adelaide.edu.au (Q.D. Nguyen).
0016-2361/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2010.09.044
839
Olive oil
Peanut oil
Canola oil
Soybean oil
Sunflower oil
Corn oil
Diesel
0.08
0.06
Viscosity (Pas)
0.04
0.02
0
290
300
310
320
330
340
350
360
Temperature (K)
The vegetable oils used were commercial food-grade canola,
corn, olive, peanut, soybean, and sunower oils, purchased from
a supermarket. Table 1 shows the fatty acid compositions of the
oils as supplied by the manufacturers. Each of the oils was blended
with No. 2 grade automotive diesel fuel to make four mixtures containing 20%, 40%, 60% and 80% oil by volume. Blending was carried
out at room temperature by adding a measured volume of the oil to
the diesel under continuous agitation with a magnetic stirrer. The
rheological properties of the oils, diesel fuel and their blends were
determined using a rate-controlled rotational rheometer (Haake
Viscotester VT550, Germany) equipped with the concentric cylinder geometry. Rheological measurements were performed out at
constant temperatures ranging from 20 C to 80 C. The sample
temperature was maintained constant to within 0.1 C by means
of a water bath (Thermoline Scientic Unistat, Australia) which circulated water and through the jacket of the rheometer. Specic
gravity of each sample was also determined as a function of temperature using a standard glass hydrometer (Stenwick Scientic,
Australia). All measurements at each temperature were repeated
at least three times and the results were averaged for subsequent
analyses. The maximum uncertainties in both viscosity and density
determinations were less than 1%.
3. Results and discussion
3.1. Viscosities of pure vegetable oils and diesel
Rheological measurements over a shear rate range of 1600 s1
indicated that the vegetable oils, the diesel fuel and their blends all
exhibit time-independent Newtonian ow behaviour in the temperature range investigated (2080 C). In Fig. 1 are shown viscosity data obtained for the pure vegetable oils and diesel fuel studied
as a function of temperature. At a given temperature, olive oil has
the highest viscosity, followed by peanut, canola, soybean, and
sunower oil with corn oil being the least viscous. Referring to
the fatty acid compositions of the oils in Table 1, the oil viscosity
Table 1
Fatty acids composition of the vegetable oils tested.
Oil
Canola
Corn
Olive
Peanut
Soybean
Sunower
Mono-unsaturated
Poly-unsaturated
8.7
13.8
17.0
18.9
14.3
11.2
60.4
25.2
69.9
45.9
24.3
26.3
30.0
61.1
13.0
35.1
61.4
62.6
Fig. 1. Viscositytemperature data for pure vegetable oils and diesel fuel.
l A exp
E
RT
1
where l is dynamic viscosity, T is absolute temperature, A is a constant, E is enthalpy of activation and R is the universal gas constant.
Values of the parameters A and E determined by least square regression together with the data tting statistics are summarized in Table 2. The goodness of t is realized with linear correlation
Table 2
Values of the constants A and E for Eq. (1) and data tting statistics for the vegetable
oils and diesel fuel.
Oil
A (106 Pa s)
E (kJ/mol)
r2
MREa (%)
Olive oil
Peanut oil
Canola oil
Soybean oil
Sunower oil
Corn oil
Diesel
0.889
1.383
1.310
1.616
0.979
1.097
7.290
27.719
26.563
26.522
25.649
26.796
26.380
15.755
0.9969
0.9938
0.9971
0.9964
0.9974
0.9995
0.9980
3.13
4.17
2.98
3.13
2.87
1.27
1.40
a
MRE: Mean relative error = 100
data points.
PN
i
840
-6
0.06
Olive Oil
Peanut oil
Canola oil
Soybean oil
Sunflower oil
Corn oil
(a)
Am (10 Pas)
Viscosity (Pas)
10
100% oil
80% oil
60% oil
40% oil
20% oil
0% oil
Model fit
0.08
0.04
0.02
0
280
290
300
310
320
330
340
350
360
0.2
Viscosity (Pas)
0.6
0.8
30
(b)
Em (kJ/mol)
0.1
100% oil
80% oil
60% oil
40% oil
20% oil
0% oil
0.4
Temperature (K)
20
Olive Oil
Peanut oil
Canola oil
Soybean oil
Sunflower oil
Corn oil
10
0.01
0
0
0.2
0.4
0.6
0.8
0.001
0.0027
0.0029
0.0031
0.0033
0.0035
1/Temperature (1/K)
Fig. 3. Arrhenius plot of viscosity versus 1/T for soybean oildiesel blends to test
the applicability of Eq. (2).
lm Am exp
Em
RT
2
where lm is the blend viscosity, and Am and Em are the corresponding model parameters for each blend at a xed composition. The
applicability of Eq. (2) to tting the viscositytemperature data
for all vegetable oildiesel mixtures at all concentrations tested is
demonstrated in Fig. 3, using the data for the soybean oildiesel
blends. From these linear plots of ln(lm) versus 1/T for all vegetable
oildiesel blends, values of Am and Em have been determined by
least square regression, and the results are presented in Fig. 4 as
function of oil concentration. It is found that Am decreases exponentially with oil concentration and Em increases linearly with volume
fraction of the oil in the blend. These results suggest that Eq. (2) can
be modied to include the effect of oil concentration on viscosity as
follows:
lm a b/ exp
c/ d
RT
841
a (106 Pa s)
b ()
c (kJ/mol)
d (kJ/mol)
MRE (%)
Canola
Corn
Olive
Peanut
Soybean
Sunower
7.980
6.344
8.675
7.728
7.534
8.120
0.166
0.161
0.117
0.165
0.229
0.124
10.939
10.453
12.032
11.105
9.834
11.251
15.551
16.128
15.254
15.634
15.646
15.477
2.59
2.09
3.20
2.62
2.87
3.58
6.19
5.11
7.50
7.52
6.42
4.69
lnlm
N
X
xi lnli
N X
N
X
i
xi xj Iij
where lm is viscosity of the mixture, li is viscosity of a pure component i of concentration xi, (in mass, mol or volume fraction), N
is the number of components, and Iij is a parameter accounting
for binary interactions between two different components (Iij = 0
for i = j).
The mixture model was tested by plotting the logarithm of
blend viscosity measured at constant temperatures against oil volume fraction. An example is illustrated in Fig. 5 with experimental
data for the canola oildiesel blends. Over the whole range of blend
compositions and temperatures studied, the logarithm of blend
viscosity increases monotonically and linearly with volume fraction of the oil in the blend. Similar results were obtained for blends
containing other types of vegetable oils investigated, with the linear correlation coefcient (r2) greater than 0.999. The linearity of
the semi-logarithmic plots of viscosity versus oil volume fraction
indicates that the interaction terms Iij in the blending rule, Eq.
(4), can be neglected for mixtures of vegetable oils and diesel fuel
studied. This is reasonable for these components of the blend
which are similar in chemical structures and form miscible solutions at all proportions within the temperature range investigated.
Similar observations have been reported by Joshi and Pegg [17] for
diesel blends with sh oil biodiesel, and Benjumea et al. [18] with
blends of palm oil biodiesel and petroleum diesel.
Thus the Arrhenius mixing rule can be applied to correlate viscosity of vegetable oildiesel blends with oil composition at a xed
temperature as follows:
ln lm ln lO ln lD / ln lD / ln lO 1 / ln lD
or
lm l/O l1/
D
0.1
Viscosity (Pa.s)
can predict blend viscosity with MRE ranging from 2.09% with the
corn oildiesel blends to 3.58% with the blends containing the sunower oil. An example of the excellent comparison between experimental data and model predictions by means of Eq. (3) is shown in
Fig. 2 for the blends containing the soybean oil. Overall, the maximum error involved in using Eq. (6) was less than 7.6% for all blends
of the six oils in diesel fuel studied, based on a population of 252
data points.
Although the empirical model represented by Eq. (3) is satisfactory for correlating the blend viscosity to composition as a function
of temperature, its application may be limited to the specic types
of vegetable oils and diesel fuel investigated. This model also requires four empirical constants which must be determined for each
oil and diesel combination from measurements of blend viscosity
at different compositions and temperatures. A more universal approach is required to develop a general model that is applicable
to diesel based blends with any vegetable oil. In this approach,
we employ the Arrhenius mixing rule that has been used for predicting transport properties of ideal liquid mixtures and solutions
based on properties of the individual components [15]. A simple
general form of the model in terms of the dynamic viscosity may
be expressed as follows [16]:
0.01
Temp (K)
293
303
313
323
333
343
353
0.001
0
0.2
0.4
0.6
0.8
lm AD
AO
AD
/
exp
EO ED / ED
RT
842
0.1
950
100% oil
80% oil
60% oil
40% oil
20% oil
0% oil
Mixture model
0.06
100% oil
80% oil
60% oil
40% oil
20% oil
0% oil
900
3
Density (kg/m )
Viscosity (Pas)
0.08
0.04
850
800
0.02
0
280
290
300
310
320
330
340
350
360
750
280
290
Temperature (K)
310
320
330
340
350
360
Temperature (K)
0.1
300
Table 4
Values of the constants a and b of Eq. (7) for the pure components of the vegetable
oilsdiesel blends.
0.01
Pure component
a (kg/m3)
b (kg/m3 K)
r2
Canola oil
Corn oil
Olive oil
Peanut oil
Soybean oil
Sunower oil
Diesel
1204.8
1240.6
1201.8
1187.7
1197.0
1239.8
1112.7
0.991
1.094
0.989
0.954
0.959
1.098
0.956
0.9980
0.9988
0.9976
0.9945
0.9978
0.9985
0.9972
0.01
0.1
temperature by using only the knowledge of the temperaturedependent viscosities of the individual pure components. As an
example, Fig. 6 demonstrates an excellent agreement between viscosities predicted using the mixture model and experimental viscosity data for the olive oildiesel blends over the whole range of
concentrations and temperatures studied. A comparison of data
for all vegetable oildiesel fuel mixtures investigated (Fig. 7) indicate that the mixture model can predict blend viscosity accurately
with a maximum error of less than 5.5%.
To be useful for predicting viscosity of vegetable oildiesel
blends, Eq. (6) requires availability of the parameters, A and E,
for the pure oil and diesel components in the blend. As shown in
Table 2, the parameter Ao for the pure oil is dependent on the type
and source of the oil, while the activation energy Eo is almost
insensitive to the oil type. Thus, in the absence of temperature
dependence data for the pure vegetable oil, the average Eo value together with one viscosity datum at a known temperature for the
pure oil can be used with Eq. (6) to predict vegetable oildiesel
blend viscosity at any composition and temperature.
m /s)
-6
100% oil
80% oil
60% oil
40% oil
20% oil
100% Diesel
Mixture model
80
4. Conclusion
60
40
20
0
290
300
310
320
330
340
350
360
Temperature (K)
Fig. 9. Comparison between experimentally measured kinematic viscosity and
kinematic viscosity predicted using the mixture model, Eq. (8), for the peanut oil
diesel blends.
mm
/aO bO T 1 /aD bD T
qm
lm
qO =qD /
lO
qm qO =qD 1/ 1 qO
Fq; /; Tm/O mD1/
mm
/
lD
qD
843
1/
9